|Publication number||US6059840 A|
|Application number||US 08/819,382|
|Publication date||May 9, 2000|
|Filing date||Mar 17, 1997|
|Priority date||Mar 17, 1997|
|Publication number||08819382, 819382, US 6059840 A, US 6059840A, US-A-6059840, US6059840 A, US6059840A|
|Inventors||Cliff N. Click, Jr.|
|Original Assignee||Motorola, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (18), Non-Patent Citations (22), Referenced by (100), Classifications (4), Legal Events (8)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates to computer language compiler methods, and more particularly to optimization of resultant performance.
Scheduling, or instruction rearrangement, has long been a part of optimizing code. It typically involves rearranging instruction order to increase processor performance. Scheduling has the secondary effect of altering which registers can be used without changing the program behavior. Calling conventions require values to be in specific registers. If the instruction that produces a subroutine parameter cannot write directly to the required register, a COPY instruction is needed just prior to the subroutine call. This COPY instruction increases code size and therefore decreases performance. Scheduling can move the defining instruction to just prior to the subroutine, produce the result directly in the required register and avoid the COPY instruction.
Note that expected savings do not come from avoiding the execution of the COPY instruction. This is because a COPY instruction itself uses few CPU resources. Most modem CPUs can issue two integer operations per cycle and the COPY instruction will only have a one-cycle latency. Further, the COPY instruction's one input will have executed many cycles previously (as there has to be an intervening subroutine call or the required register will be legal and the COPY instruction can be avoided).
For larger integer-intensive codes like operating systems, data bases, compilers, and complex GUIs, subroutine calls are a large part of the static instruction count for these programs. For example, with respect to many such large integer-intensive codes, about 10% of the static instruction count for these codes will be COPY instructions. The large number of COPY instructions generated uses valuable space in the instruction cache. This in turn causes I-cache misses which are very expensive. Some of these generated COPY instructions can be avoided by rescheduling.
Scheduling has a long history with much research. This research has been focused on raw performance, generally considering only instruction latency and available parallelism (i.e., superscalar or VLIW). Simplifying assumptions utilized during scheduling include assuming all instructions are in I-cache and that subroutine calls are not that interesting. This later is because subroutine calls typically contain far more work than is being saved by good scheduling in the one block containing the call. Coalescing COPY instructions has not been a focus of this work.
COPY instruction coalescing has typically been done either as a separate compiler pass (called Copy Propagation or Copy Forwarding) or as part of the register allocater in a compiler. Both of these techniques attempt to change any registers defined by a COPY instruction to be the same registers that the COPY instruction is using. If successful, the COPY instruction will move a register onto itself, and can be easily removed. These techniques are global since they can change register assignments across a whole program. However, these techniques do not reorder instructions. If a value is defined before a subroutine CALL, is required in a specific register after the subroutine is entered, and the subroutine "clobbers" that register, then these techniques will require a COPY instruction. This invention attempts to move defining instructions past subroutine CALLs so that their values can be defined directly into the required registers. FIG. 3 below is an example of a COPY instruction that cannot be removed by prior art, but can be removed by this invention.
FIGS. 1 and 2 are dataflow charts used to illustrate COPY instruction elimination and Register Allocation techniques for eliminating COPY instructions. FIG. 1 is an example where COPY Propagation can successfully remove a COPY instruction. Instruction 70 defines register RA. In the example, an ADD instruction is shown as instruction 70. However, this technique works with any instruction that defines register RA. Instruction 72 copies register RA to register RB. Instruction 74 utilizes register RB. In this instance, an ADD instruction is shown as instruction 74 that adds register RB to another register. However, this technique works other instructions than ADD instructions.
COPY elimination operates by looking at a COPY instruction such as shown in instruction 72, and determining whether the source register in COPY instruction 72 is clobbered between instruction 72 and the instruction that uses the COPY instruction result in instruction 74. In the example in FIG. 1, register RA is not clobbered between instruction 72 where it is copied to register RB, and instruction 74 where register RB is utilized. In this example, there is no other use for register RB, and COPY instruction 72 therefore becomes unused. Unused or "Dead" COPY instructions can therefore be eliminated. The resulting code is shown on the right hand side of FIG. 1, where instruction 70 stays the same, generating an output value in register RA. Instruction 74 is modified to replace register RB with register RA resulting in instruction 76. COPY instruction 72 has been eliminated.
Register Allocation generates the same results as COPY instruction elimination in this example, but with a different algorithm. This technique looks for a "live range" for register RA, and a live range for register RB. If the live ranges for register RA and register RB do not interfere, then the two registers can be coalesced. Thus, in the FIG. 1 example, the live range of register RA is between instruction 70 and instruction 72, and the live range of register RB is between instruction 72 and instruction 74. Since there is no interference between the two live ranges, the registers can be coalesced, and the COPY instruction 72 can be eliminated.
The way that COPY instruction coalescing works is that a single combined register RAB is created, replacing registers RA and RB. This turns COPY instruction 72 into a COPY of register RAB into itself, which can be easily eliminated.
FIG. 2 is an example where a COPY Propagation in a forward direction fails to eliminate COPY instructions, but Register Allocation does remove COPY instructions. Register RA is defined in instruction 80. In this example, an ADD instruction is shown, wherein the result is placed in register RA. However, this technique works with other instructions. The resulting value in register RA is copied to register R3 in instruction 82. This resulting value in register R3 is utilized in the call to subroutine FOO in instruction 84. Note that the use of register R3 is implicit here. Whenever subroutine CALLs are generated by higher level language compilers, calling conventions are used. These conventions dictate what registers are used to communicate between the calling and called routines, and what registers are clobbered by the called routine.
In this example, COPY Propagation fails because resulting register RB in FIG. 1 cannot be renamed. This is because register R3 is fixed by the calling convention. In this instance, R3 is shown, since this is part of the PowerPC™ calling convention. However, other computer architectures utilize other calling conventions. COPY elimination fails here because it cannot rename register R3 to register RA, due to the calling convention. Note, that COPY Propagation has been shown, which works in the forward direction. Whereas in this example COPY Propagation operating in the backward direction would eliminate the COPY instruction by defining RA instruction in instruction 80 to be R3, a similar counter example is easily generated that fails in the backward direction.
Register Allocation is similar in operation to that shown in FIG. 1. A live range for register RA is identified between instructions 80 and 82. Likewise, a live range is identified for register R3 between instructions 82 and 84. Since the two live ranges do not interfere, the two registers can be coalesced, a combined register RA3 can be utilized in instructions 82 and 84, resulting in a copy of RA3 to register RA3 in instruction 82, which can be easily eliminated. The result is that instruction 80 with the result in RA is replaced by instruction 86 which is identical except that its result is left in register R3. The result left in R3 can then be used correctly by the call to subroutine FOO in instruction 84.
FIG. 3 is a dataflow chart illustrating an example where COPY instructions cannot be eliminated by either COPY Propagation or register allocation. Register RA is defined in instruction 90. As with step 70, an ADD instruction is shown in instruction 90. However, this technique will work with other instructions. Subroutine FOO is called in instruction 92. This has the result of clobbering all the registers in the calling convention. This is followed by a copy of register RA to register R3 in instruction 94. Finally, register R3 is used in a call to BAR in instruction 96.
COPY Propagation fails with the FIG. 3 example for the same reason that it failed with the FIG. 2 example, which is that register R3 cannot be renamed, due to its being part of the calling convention. Register allocation fails in this example because the live range for register RA interferes with the live range for register R3. This is because the live range for register RA is between instructions 90 and 94 but the live range of R3 is not just between instructions 94 and 96, but also includes the call to subroutine FOO at instruction 92. This is because register R3 is clobbered by the calling convention in the call to subroutine FOO at instruction 92. Due to this interference between the live ranges of the two registers, register allocation cannot coalesce register R3 and RA.
It would be advantageous to be able to remove COPY instructions in instances where neither COPY Propagation nor Register Allocation is effective in removing such COPY instructions.
A better understanding of the present invention can be obtained when the following detailed description is considered in conjunction with the following drawings, in which:
FIG. 1 illustrates, in data flow chart format, a prior art example of compiler optimization using COPY Propagation;
FIG. 2 illustrates, in data flow chart format, a prior art example of compiler optimization using register coalescing;
FIG. 3 illustrates, in flow chart form, instruction scheduling according to one embodiment of the present invention;
FIG. 4 illustrates, in data flow chart format, compiler optimization using instruction scheduling according to one embodiment of the present invention;
FIG. 5 illustrates, in block diagram form, a computer system according to one embodiment of the present invention; and
FIG. 6 is a block diagram showing a General Purpose Computer used to implement the present invention.
This invention provides a method for effectively eliminating COPY instructions in call-intensive codes by rearranging instructions. These COPY instructions would otherwise be required by the calling conventions. Removing the COPY instructions reduces code size, and therefore I-cache miss rate.
The method schedules instructions with the intent of removing COPY instructions. It can be used in conjunction with any prior scheduling technique. The method has three primary parts.
1. Build def-use chains.
2. Coalesce all COPY instructions creating an illegal or unschedulable program.
3. Schedule the instructions, preserving the def-use chains as much as possible.
Def-use chains represent program as directed graphs. Nodes in the graphs are program instructions. Arcs in the graphs go from the instruction defining some value to the instruction using that value. In the final programs each arc will be associated with a physical register. During scheduling, arcs may not be associated with a physical register; equivalently, arcs can be associated with virtual registers drawn from an infinite pool of registers. Either the defining instruction or the using instruction can require its value to be in a particular physical register. It is possible that both instructions require different physical registers. In this situation, a COPY instruction is required to be generated. Def-use chains are built using standard techniques for basic block schedulers.
Copy instructions are coalesced by combining a COPY instruction's input and output arcs. The final arc may require a particular physical register on the arc's head or tail. If the arc is unconstrained, then the COPY instruction is trivially removed. If the arc is constrained to a single register, then it may be possible to schedule the code so that the register is not clobbered between the defining instruction and the using instruction. If rescheduled, then the COPY instruction can be removed. If not, then the COPY instruction must remain. If both ends of the arc require different physical registers, a COPY instruction is still required.
Coalescing COPY instructions during scheduling is one key idea of this invention. Coalescing COPY instructions in a compiler is typically done either by the register allocater or by a separate compiler pass. In either case, another compiler pass is unable to rearrange instructions.
The remaining phase of the invention deals with the consequences of coalescing COPY instructions. A typical list scheduler is used. List schedulers maintain a list of "ready" instructions; instructions whose inputs have already been scheduled. The scheduler pulls the best instruction from a "ready list" containing the ready instructions using some heuristic, then schedules it. Scheduling an instruction typically makes more instructions ready. The process repeats until all instructions have been scheduled, and the ready list is empty.
The art of list scheduling involves selecting a good heuristic. The heuristic described herein is designed to produce a legal schedule if one is possible. The heuristic follows these basic steps:
1. Find the next subroutine CALL or end-of-block (using a simple linear scan).
2. Find the set of instructions that must be scheduled before this subroutine CALL due to data dependencies (by walking the def-use chains already present).
3. Schedule this set of instructions, then the subroutine CALL, then repeat until all instructions have been scheduled.
If the ready instruction selected to be scheduled next defines a register that is currently in use, then there is a conflict. Resolving the conflict usually requires a COPY instruction to be generated, where the value in use is copied into virtual register and the value's users are altered to reflect this. Since conflicts imply generation of COPY instructions, the heuristic tries to avoid conflicts. Scheduling the subset of ready instructions required by the next subroutine CALL is further refined by more heuristics:
1. Ready instructions defining virtual registers are scheduled before ones defining physical registers. Virtual registers never conflict.
2. Instructions that do not conflict and use many physical registers are selected next. When the last use of a physical register is scheduled, that register no longer causes conflicts.
3. After all instructions are scheduled, a separate pass can reschedule the instructions with the intent of reducing latency.
FIG. 4 is a flow chart that illustrates removing COPY instructions utilizing rescheduling. First, Def-Use chains are built, step 52. Def-Use chains are described earlier and are built using standard techniques. Next, COPY instructions are coalesced, step 54, as described earlier. Next a list of instructions whose inputs are all available is generated. This ready list is built in step 56.
A loop is entered in step 58. While the ready list is not empty, the loop iterates. When the ready list is finally empty, the loop exits and the method is done. Within the loop, an instruction is first selected from the ready list, in step 62. In step 64, if the selected instruction conflicts with any values that are currently live, then the conflict is resolved. A conflict occurs if a selected instructions writes to a physical register which currently holds a value needed by some other instruction. Resolving such a conflict requires inserting a COPY instructions to move the value in the physical register to a virtual register. Next the selected instruction is scheduled in the basic block, step 66. Scheduling an instruction makes its inputs available for other instructions. If any of these other instructions now have all their inputs available, they are added to the ready list, step 68. Finally the loop repeats starting at step 58.
FIG. 5 is a dataflow chart that illustrates the method shown in FIG. 4 applied to the example shown in FIG. 3. In this example, the Def-Use chains are shown explicitly by arcs. Coalescing COPY instructions in step 54 replaces register RA in instruction 90 with the coalesced register R3, and the COPY instruction 94 is removed. The result is shown on the right hand side of FIG. 5. Next, instructions 92 and 98 are ready and placed on the ready list, step 56.
Since the ready list is not empty, step 58, a next instruction is selected. Different methods can be used to select the next instruction. Preferably the following technique is used. Instruction 98 is not needed by the CALL instruction, 92, and it defines a physical register R3. Scheduling instruction 98 next would lead to a conflict with the next call. Instead, instruction 92 is scheduled. The ready list will then contain only instruction 98, and the method repeats the loop starting at step 58.
Instruction 98 now must be scheduled (since it is the only instruction on the ready list) . Since it does not conflict, no conflict resolution is done. Instruction 98 is then scheduled and instruction 96 is placed on the ready list. Finally instruction 96 is scheduled, the ready list runs dry and the method is done.
This method illustrates a technique for removing COPY instructions. The actual execution time of COPY instructions is quite small for modern microprocessors. However, in call-intensive programs, COPY instructions represent a significant percentage of all instructions and they use valuable I-cache space. I-cache misses are a significant portion of the running time of these large programs, so reducing code size reduces running time. Traditional techniques for removing COPY instructions are limited to register selection (the Copy Elimination or Register Allocation techniques described above) and do not attempt code motion. Traditional code motion techniques do not attempt to remove COPY instructions and miss many opportunities to do so.
FIG. 6 is a block diagram showing a General Purpose Computer 20. The General Purpose Computer 20 has a Computer Processor 22, and Memory 24, connected by a Bus 26. Memory 24 includes relatively high speed machine readable media such as DRAM, SRAM, ROM, FLASH, EEPROM, and bubble memory. Also connected to the Bus are Secondary Storage 30, External Storage 32, output devices such as a monitor 34, input devices such as a keyboard (with mouse) 36, and printers 38. Secondary Storage 30 includes machine readable media such as hard disk drives, magnetic drum, and bubble memory. External Storage 32 includes machine readable media such as floppy disks, removable hard drives, magnetic tape, CD-ROM, and even other computers, possibly connected via a communications line. The distinction drawn here between Secondary Storage 30 and External Storage 32 is primarily for convenience in describing the invention. As such, it should be appreciated that there is substantial functional overlap between these elements. Executable versions of computer software 33, such as a compiler and user programs generated by such a compiler can be read from the External Storage 32 and loaded for execution directly into the Memory 34, or stored on the Secondary Storage 30 prior to loading into Memory 34 and execution.
Those skilled in the art will recognize that modifications and variations can be made without departing from the spirit of the invention. Therefore, it is intended that this invention encompass all such variations and modifications as fall within the scope of the appended claims.
Claim elements and steps herein have been numbered and/or lettered solely as an aid in readability and understanding. As such, the numbering and/or lettering in itself is not intended to and should not be taken to indicate the ordering of elements and/or steps in the claims.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4571678 *||Nov 5, 1982||Feb 18, 1986||International Business Machines Corporation||Register allocation and spilling via graph coloring|
|US5367651 *||Nov 30, 1992||Nov 22, 1994||Intel Corporation||Integrated register allocation, instruction scheduling, instruction reduction and loop unrolling|
|US5367696 *||Oct 1, 1991||Nov 22, 1994||Fuji Xerox Co., Ltd.||Register allocation technique in a program translating apparatus|
|US5396631 *||Aug 31, 1993||Mar 7, 1995||Fujitsu Limited||Compiling apparatus and a compiling method|
|US5491823 *||Jan 25, 1994||Feb 13, 1996||Silicon Graphics, Inc.||Loop scheduler|
|US5530866 *||Oct 13, 1994||Jun 25, 1996||Tera Computer Company||Register allocation methods having upward pass for determining and propagating variable usage information and downward pass for binding; both passes utilizing interference graphs via coloring|
|US5761514 *||Aug 31, 1995||Jun 2, 1998||International Business Machines Corporation||Register allocation method and apparatus for truncating runaway lifetimes of program variables in a computer system|
|US5784066 *||Nov 22, 1995||Jul 21, 1998||International Business Machines Corporation||Method and apparatus for using partner information to color nodes in an interference graph within a computer system|
|US5790866 *||Feb 13, 1995||Aug 4, 1998||Kuck And Associates, Inc.||Method of analyzing definitions and uses in programs with pointers and aggregates in an optimizing compiler|
|US5790867 *||Jan 2, 1996||Aug 4, 1998||International Business Machines Corporation||Compiler with extended redundant copy elimination|
|US5835776 *||Nov 17, 1995||Nov 10, 1998||Sun Microsystems, Inc.||Method and apparatus for instruction scheduling in an optimizing compiler for minimizing overhead instructions|
|US5878261 *||May 15, 1996||Mar 2, 1999||Hewlett-Packard Company||Method for restructuring code to reduce procedure call overhead|
|US5901317 *||Mar 25, 1996||May 4, 1999||Sun Microsystems, Inc.||Method and system for register allocation using multiple interference graphs|
|US5903761 *||Oct 31, 1997||May 11, 1999||Preemptive Solutions, Inc.||Method of reducing the number of instructions in a program code sequence|
|US5923883 *||Mar 12, 1997||Jul 13, 1999||Matsushita Electric Industrial Co., Ltd.||Optimization apparatus which removes transfer instructions by a global analysis of equivalence relations|
|US5930507 *||May 30, 1996||Jul 27, 1999||Fujitsu Limited||Compiling processing apparatus|
|US5937196 *||Sep 19, 1997||Aug 10, 1999||International Business Machines Corporation||Compiling with partial copy propagation|
|US5940621 *||Jan 23, 1996||Aug 17, 1999||Hewlett-Packard Company||Language independent optimal size-based storage allocation|
|1||Alfred V. Aho, Ravi Sethi and Jeffrey D. Ullman "Compilers Principles, Techniques,and Tools", Addison-Wesley Chapter 10, Mar. 1988.|
|2||*||Alfred V. Aho, Ravi Sethi and Jeffrey D. Ullman Compilers Principles, Techniques,and Tools , Addison Wesley Chapter 10, Mar. 1988.|
|3||Chow, F.; "Minimizing Register Usage Penalty at Procedure Calls"; Proceedings of the SIGPLAN '88 Conference on Programming Language Design and Implementation; pp. 85-94, Jun. 1998.|
|4||*||Chow, F.; Minimizing Register Usage Penalty at Procedure Calls ; Proceedings of the SIGPLAN 88 Conference on Programming Language Design and Implementation; pp. 85 94, Jun. 1998.|
|5||Frederick Chow, et al., "Register Allocation by Priority-based Coloring", Proceedings of the ACM SIGPLAN '84 Symposium on Compiler Construction SIGPLAN Notices, vol. 19,No. 6, Jun. 1984, pp. 222-232.|
|6||*||Frederick Chow, et al., Register Allocation by Priority based Coloring , Proceedings of the ACM SIGPLAN 84 Symposium on Compiler Construction SIGPLAN Notices, vol. 19,No. 6, Jun. 1984, pp. 222 232.|
|7||George, L.; Appel, A.; "Iterated Register Coalescing"; ACM Transactions on Programming Languages and Systems; vol. 18, No. 3, pp. 300-324, Mar. 1996.|
|8||*||George, L.; Appel, A.; Iterated Register Coalescing ; ACM Transactions on Programming Languages and Systems; vol. 18, No. 3, pp. 300 324, Mar. 1996.|
|9||Gregory J. Chaitin, et al, "Register Allocation Via Coloring", IBM T. J. Watson Research Center, RC 8395 (#36543) Aug. 4, 1980, Computer Science, pp. 1-12.|
|10||*||Gregory J. Chaitin, et al, Register Allocation Via Coloring , IBM T. J. Watson Research Center, RC 8395 ( 36543) Aug. 4, 1980, Computer Science, pp. 1 12.|
|11||John Hennessy, et al., "Postpass Code Optimization of Pipeline Constraints", ACM Transaction on Programming Languages and Systems, vol. 5, No. 3, Jul. 1983, pp. 422-448.|
|12||*||John Hennessy, et al., Postpass Code Optimization of Pipeline Constraints , ACM Transaction on Programming Languages and Systems, vol. 5, No. 3, Jul. 1983, pp. 422 448.|
|13||Joseph A. Fisher, "Trace Scheduling: A Technique for Global Microcode Compaction", IEEE Transactions on Computers, vol. C-30., No. 7, Jul. 1981, pp. 478-490.|
|14||*||Joseph A. Fisher, Trace Scheduling: A Technique for Global Microcode Compaction , IEEE Transactions on Computers, vol. C 30., No. 7, Jul. 1981, pp. 478 490.|
|15||Monica Lam, "Software Pipelining: An Effective Scheduling Techique for VLIW Machines", Proceedings of SIGPLAN '88 Conf. on Programming Language Design/Implementation, Atlanta, GA, Jun. 22-24, 1988, pp. 318-327.|
|16||*||Monica Lam, Software Pipelining: An Effective Scheduling Techique for VLIW Machines , Proceedings of SIGPLAN 88 Conf. on Programming Language Design/Implementation, Atlanta, GA, Jun. 22 24, 1988, pp. 318 327.|
|17||*||Park, J.; Moon, S. M.; Optimistic Register Coalescing ; Proceedings of the 1998 International Conference on Parallel Architectures and Compilation Techniques; pp. 196 204, Oct. 1998.|
|18||Park, J.; Moon, S.-M.; "Optimistic Register Coalescing"; Proceedings of the 1998 International Conference on Parallel Architectures and Compilation Techniques; pp. 196-204, Oct. 1998.|
|19||Phillip B. Gibbons, et al., "Efficient Instruction Scheduling for a Pipelined Architecture", Proceedings of SIGPLAN '88 Conf. on Compiler Construction SIGPLAN Notices 21(7), Jul.,1988,pp. 1-3.|
|20||*||Phillip B. Gibbons, et al., Efficient Instruction Scheduling for a Pipelined Architecture , Proceedings of SIGPLAN 88 Conf. on Compiler Construction SIGPLAN Notices 21(7), Jul.,1988,pp. 1 3.|
|21||Preston Briggs, Dissertation "Register Allocation via Graph Coloring", Rice University, Houston, Texas, Apr. 1992, pp. 1-142.|
|22||*||Preston Briggs, Dissertation Register Allocation via Graph Coloring , Rice University, Houston, Texas, Apr. 1992, pp. 1 142.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US6314471||Nov 13, 1998||Nov 6, 2001||Cray Inc.||Techniques for an interrupt free operating system|
|US6321379 *||Dec 23, 1998||Nov 20, 2001||Cray Inc.||Method and system for target register allocation|
|US6353829||Dec 23, 1998||Mar 5, 2002||Cray Inc.||Method and system for memory allocation in a multiprocessing environment|
|US6415433||Dec 23, 1998||Jul 2, 2002||Cray Inc.||Method and system for identifying locations to move portions of the computer program|
|US6430676||Dec 23, 1998||Aug 6, 2002||Cray Inc.||Method and system for calculating instruction lookahead|
|US6446258 *||Nov 3, 1998||Sep 3, 2002||Intle Corporation||Interactive instruction scheduling and block ordering|
|US6457173 *||Aug 20, 1999||Sep 24, 2002||Hewlett-Packard Company||Automatic design of VLIW instruction formats|
|US6480818||Nov 13, 1998||Nov 12, 2002||Cray Inc.||Debugging techniques in a multithreaded environment|
|US6665688||Dec 23, 1998||Dec 16, 2003||Cray Inc.||Method and system for automatically regenerating data on-demand|
|US6694435||Jul 25, 2001||Feb 17, 2004||Apple Computer, Inc.||Method of obfuscating computer instruction streams|
|US6848097||Sep 18, 2002||Jan 25, 2005||Cray Inc.||Debugging techniques in a multithreaded environment|
|US6862635||Jul 27, 1999||Mar 1, 2005||Cray Inc.||Synchronization techniques in a multithreaded environment|
|US6952827||Nov 13, 1998||Oct 4, 2005||Cray Inc.||User program and operating system interface in a multithreaded environment|
|US6961925||Apr 3, 2001||Nov 1, 2005||Cray Inc.||Parallelism performance analysis based on execution trace information|
|US7000227 *||Sep 29, 2000||Feb 14, 2006||Intel Corporation||Iterative optimizing compiler|
|US7020767||Feb 23, 2001||Mar 28, 2006||Cray Inc.||Techniques for reducing the rate of instruction issuance|
|US7117330||Jun 12, 2003||Oct 3, 2006||Cray Inc.||Synchronization techniques in a multithreaded environment|
|US7165150||Oct 30, 2003||Jan 16, 2007||Cray Inc.||Restricting access to memory in a multithreaded environment|
|US7191444||Sep 16, 2003||Mar 13, 2007||Cray Inc.||Stream management in a multithreaded environment|
|US7322045||Jan 23, 2004||Jan 22, 2008||Apple Inc.||Method of obfuscating computer instruction streams|
|US7360221 *||Sep 10, 2003||Apr 15, 2008||Cray Inc.||Task swap out in a multithreaded environment|
|US7383544 *||Mar 4, 2004||Jun 3, 2008||International Business Machines Corporation||Compiler device, method, program and recording medium|
|US7392525||Oct 1, 2003||Jun 24, 2008||Cray Inc.||Inter-thread long jumps in a multithreaded environment|
|US7426732||Oct 10, 2003||Sep 16, 2008||Cray Inc.||Placing a task of a multithreaded environment in a known state|
|US7428727||Sep 9, 2004||Sep 23, 2008||Cray Inc.||Debugging techniques in a multithreaded environment|
|US7536690||Sep 16, 2003||May 19, 2009||Cray Inc.||Deferred task swapping in a multithreaded environment|
|US7558889||Oct 29, 2003||Jul 7, 2009||Cray Inc.||Accessing a collection of data items in a multithreaded environment|
|US7558910||Oct 30, 2003||Jul 7, 2009||Cray Inc.||Detecting access to a memory location in a multithreaded environment|
|US7653710||May 21, 2003||Jan 26, 2010||Qst Holdings, Llc.||Hardware task manager|
|US7660984||May 13, 2003||Feb 9, 2010||Quicksilver Technology||Method and system for achieving individualized protected space in an operating system|
|US7668229||Apr 4, 2007||Feb 23, 2010||Qst Holdings, Llc||Low I/O bandwidth method and system for implementing detection and identification of scrambling codes|
|US7739667||Oct 19, 2005||Jun 15, 2010||Cray Inc.||Parallelism performance analysis based on execution trace information|
|US7752419||Dec 12, 2001||Jul 6, 2010||Qst Holdings, Llc||Method and system for managing hardware resources to implement system functions using an adaptive computing architecture|
|US7809050||Oct 13, 2009||Oct 5, 2010||Qst Holdings, Llc||Method and system for reconfigurable channel coding|
|US7822109||Mar 28, 2003||Oct 26, 2010||Qst Holdings, Llc.||Method and system for reconfigurable channel coding|
|US7865847||Jan 25, 2008||Jan 4, 2011||Qst Holdings, Inc.||Method and system for creating and programming an adaptive computing engine|
|US7904603||Sep 10, 2009||Mar 8, 2011||Qst Holdings, Llc||Adaptable datapath for a digital processing system|
|US7904685||Jun 12, 2003||Mar 8, 2011||Cray Inc.||Synchronization techniques in a multithreaded environment|
|US7937538||May 7, 2009||May 3, 2011||Qst Holdings, Llc||External memory controller node|
|US7937539||May 7, 2009||May 3, 2011||Qst Holdings, Llc||External memory controller node|
|US7937591||Oct 25, 2002||May 3, 2011||Qst Holdings, Llc||Method and system for providing a device which can be adapted on an ongoing basis|
|US7941614||May 7, 2009||May 10, 2011||QST, Holdings, Inc||External memory controller node|
|US7979646||Oct 15, 2008||Jul 12, 2011||Qst Holdings, Inc.||External memory controller node|
|US7979853||Jan 24, 2008||Jul 12, 2011||International Business Machines Corporation||Compiler device, method, program and recording medium|
|US7984247||Oct 15, 2008||Jul 19, 2011||Qst Holdings Llc||External memory controller node|
|US8108656||Aug 29, 2002||Jan 31, 2012||Qst Holdings, Llc||Task definition for specifying resource requirements|
|US8200799||Feb 9, 2009||Jun 12, 2012||Qst Holdings Llc||Hardware task manager|
|US8225073||Mar 6, 2009||Jul 17, 2012||Qst Holdings Llc||Apparatus, system and method for configuration of adaptive integrated circuitry having heterogeneous computational elements|
|US8249135||Aug 20, 2010||Aug 21, 2012||Qst Holdings Llc||Method and system for reconfigurable channel coding|
|US8250339||Dec 21, 2007||Aug 21, 2012||Qst Holdings Llc||Apparatus, method, system and executable module for configuration and operation of adaptive integrated circuitry having fixed, application specific computational elements|
|US8266388||Jul 7, 2011||Sep 11, 2012||Qst Holdings Llc||External memory controller|
|US8266609 *||Mar 23, 2006||Sep 11, 2012||Microsoft Corporation||Efficient placement of software transactional memory operations around procedure calls|
|US8276135||Nov 7, 2002||Sep 25, 2012||Qst Holdings Llc||Profiling of software and circuit designs utilizing data operation analyses|
|US8356161||Oct 15, 2008||Jan 15, 2013||Qst Holdings Llc||Adaptive processor for performing an operation with simple and complex units each comprising configurably interconnected heterogeneous elements|
|US8380884||Mar 7, 2011||Feb 19, 2013||Altera Corporation||Adaptable datapath for a digital processing system|
|US8442096||Jul 8, 2009||May 14, 2013||Qst Holdings Llc||Low I/O bandwidth method and system for implementing detection and identification of scrambling codes|
|US8533431||Oct 15, 2008||Sep 10, 2013||Altera Corporation||Adaptive integrated circuitry with heterogeneous and reconfigurable matrices of diverse and adaptive computational units having fixed, application specific computational elements|
|US8543794||Jan 19, 2012||Sep 24, 2013||Altera Corporation||Adaptive integrated circuitry with heterogenous and reconfigurable matrices of diverse and adaptive computational units having fixed, application specific computational elements|
|US8543795||Jan 19, 2012||Sep 24, 2013||Altera Corporation||Adaptive integrated circuitry with heterogeneous and reconfigurable matrices of diverse and adaptive computational units having fixed, application specific computational elements|
|US8589660||May 24, 2010||Nov 19, 2013||Altera Corporation||Method and system for managing hardware resources to implement system functions using an adaptive computing architecture|
|US8607211||Apr 27, 2012||Dec 10, 2013||International Business Machines Corporation||Linking code for an enhanced application binary interface (ABI) with decode time instruction optimization|
|US8612959||Oct 3, 2011||Dec 17, 2013||International Business Machines Corporation||Linking code for an enhanced application binary interface (ABI) with decode time instruction optimization|
|US8615745||Oct 3, 2011||Dec 24, 2013||International Business Machines Corporation||Compiling code for an enhanced application binary interface (ABI) with decode time instruction optimization|
|US8615746||Apr 30, 2012||Dec 24, 2013||International Business Machines Corporation||Compiling code for an enhanced application binary interface (ABI) with decode time instruction optimization|
|US8706916||Feb 15, 2013||Apr 22, 2014||Altera Corporation||Adaptable datapath for a digital processing system|
|US8756591||Oct 3, 2011||Jun 17, 2014||International Business Machines Corporation||Generating compiled code that indicates register liveness|
|US8767804||Aug 20, 2012||Jul 1, 2014||Qst Holdings Llc||Method and system for reconfigurable channel coding|
|US8769214||Sep 11, 2012||Jul 1, 2014||Qst Holdings Llc||External memory controller node|
|US8782196||Jun 11, 2012||Jul 15, 2014||Sviral, Inc.||Hardware task manager|
|US8799882||Mar 23, 2006||Aug 5, 2014||Microsoft Corporation||Compiler support for optimizing decomposed software transactional memory operations|
|US8880849||Aug 20, 2012||Nov 4, 2014||Altera Corporation||Apparatus, method, system and executable module for configuration and operation of adaptive integrated circuitry having fixed, application specific computational elements|
|US9002998||Aug 6, 2013||Apr 7, 2015||Altera Corporation||Apparatus and method for adaptive multimedia reception and transmission in communication environments|
|US9015352||Mar 31, 2014||Apr 21, 2015||Altera Corporation||Adaptable datapath for a digital processing system|
|US9037834||Nov 18, 2013||May 19, 2015||Altera Corporation||Method and system for managing hardware resources to implement system functions using an adaptive computing architecture|
|US9164952||Sep 24, 2013||Oct 20, 2015||Altera Corporation||Adaptive integrated circuitry with heterogeneous and reconfigurable matrices of diverse and adaptive computational units having fixed, application specific computational elements|
|US9330058||Aug 7, 2014||May 3, 2016||Altera Corporation||Apparatus, method, system and executable module for configuration and operation of adaptive integrated circuitry having fixed, application specific computational elements|
|US9396161||May 18, 2015||Jul 19, 2016||Altera Corporation|
|US9594723||Mar 13, 2013||Mar 14, 2017||Altera Corporation||Apparatus, system and method for configuration of adaptive integrated circuitry having fixed, application specific computational elements|
|US9665397||Jul 15, 2014||May 30, 2017||Cornami, Inc.||Hardware task manager|
|US20020038332 *||Feb 23, 2001||Mar 28, 2002||Alverson Gail A.||Techniques for an interrupt free operating system|
|US20020129339 *||Apr 3, 2001||Sep 12, 2002||Callahan Charles David||Parallelism performance analysis based on execution trace information|
|US20020184291 *||May 31, 2001||Dec 5, 2002||Hogenauer Eugene B.||Method and system for scheduling in an adaptable computing engine|
|US20040003383 *||Jun 28, 2002||Jan 1, 2004||Microsoft Corporation||Stripping of unnecessary information from source code|
|US20040064816 *||Oct 1, 2003||Apr 1, 2004||Alverson Gail A.||Inter-thread long jumps in a multithreaded environment|
|US20040064818 *||Sep 16, 2003||Apr 1, 2004||Alverson Gail A.||Deferred task swapping in a multithreaded environment|
|US20040078795 *||Oct 10, 2003||Apr 22, 2004||Alverson Gail A.||Placing a task of a multithreaded environment in a known state|
|US20040088711 *||Sep 10, 2003||May 6, 2004||Alverson Gail A.||Task swap out in a multithreaded environment|
|US20040093603 *||Sep 16, 2003||May 13, 2004||Alverson Gail A.||Stream management in a multithreaded environment|
|US20040093605 *||Oct 29, 2003||May 13, 2004||Alverson Gail A.||Accessing a collection of data items in a multithreaded environment|
|US20040098721 *||Oct 30, 2003||May 20, 2004||Alverson Gail A.||Restricting access to memory in a multithreaded environment|
|US20040237075 *||Mar 4, 2004||Nov 25, 2004||Motohiro Kawahito||Compiler device, method, program and recording medium|
|US20050021898 *||Oct 30, 2003||Jan 27, 2005||Alverson Gail A.||Detecting access to a memory location in a multithreaded environment|
|US20050034024 *||Sep 9, 2004||Feb 10, 2005||Alverson Gail A.||Debugging techniques in a multithreaded environment|
|US20060101416 *||Oct 19, 2005||May 11, 2006||Callahan Charles D Ii||Parallelism performance analysis based on execution trace information|
|US20060129997 *||Dec 13, 2004||Jun 15, 2006||Stichnoth James M||Optimized layout for managed runtime environment|
|US20060200811 *||Mar 7, 2006||Sep 7, 2006||Cheng Stephen M||Method of generating optimised stack code|
|US20070169030 *||Mar 23, 2006||Jul 19, 2007||Microsoft Corporation||Compiler support for optimizing decomposed software transactional memory operations|
|US20070169031 *||Mar 23, 2006||Jul 19, 2007||Microsoft Corporation||Efficient placement of software transactional memory operations around procedure calls|
|US20080184213 *||Jan 24, 2008||Jul 31, 2008||International Business Machines Corporation||Compiler device, method, program and recording medium|
|USRE42743||May 15, 2008||Sep 27, 2011||Qst Holdings, Llc||System for authorizing functionality in adaptable hardware devices|
|Mar 17, 1997||AS||Assignment|
Owner name: MOTOROLA, INC., ILLINOIS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:CLICK, CLIFF N. JR.;REEL/FRAME:008457/0952
Effective date: 19970312
|Sep 26, 2003||FPAY||Fee payment|
Year of fee payment: 4
|May 7, 2004||AS||Assignment|
Owner name: FREESCALE SEMICONDUCTOR, INC., TEXAS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:MOTOROLA, INC.;REEL/FRAME:015698/0657
Effective date: 20040404
Owner name: FREESCALE SEMICONDUCTOR, INC.,TEXAS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:MOTOROLA, INC.;REEL/FRAME:015698/0657
Effective date: 20040404
|Feb 2, 2007||AS||Assignment|
Owner name: CITIBANK, N.A. AS COLLATERAL AGENT, NEW YORK
Free format text: SECURITY AGREEMENT;ASSIGNORS:FREESCALE SEMICONDUCTOR, INC.;FREESCALE ACQUISITION CORPORATION;FREESCALE ACQUISITION HOLDINGS CORP.;AND OTHERS;REEL/FRAME:018855/0129
Effective date: 20061201
Owner name: CITIBANK, N.A. AS COLLATERAL AGENT,NEW YORK
Free format text: SECURITY AGREEMENT;ASSIGNORS:FREESCALE SEMICONDUCTOR, INC.;FREESCALE ACQUISITION CORPORATION;FREESCALE ACQUISITION HOLDINGS CORP.;AND OTHERS;REEL/FRAME:018855/0129
Effective date: 20061201
|Nov 19, 2007||REMI||Maintenance fee reminder mailed|
|May 9, 2008||LAPS||Lapse for failure to pay maintenance fees|
|Jul 1, 2008||FP||Expired due to failure to pay maintenance fee|
Effective date: 20080509
|Dec 21, 2015||AS||Assignment|
Owner name: FREESCALE SEMICONDUCTOR, INC., TEXAS
Free format text: PATENT RELEASE;ASSIGNOR:CITIBANK, N.A., AS COLLATERAL AGENT;REEL/FRAME:037354/0225
Effective date: 20151207